Adaptive Neural Compression of Biometric Data in Health Game – Experimental Compression Using AI on Real-Time Vitals
Shah Abhishek Satyendra
Chikitsak Samuha’s Sir Sitaram and Lady Shantabai Patkar College of Arts and Science and V.P. Varde College of Commerce and Economics, Mumbai, India
Abstract :
In this case we present a variable adaptive neural solution which reduces the bandwidth and storage requirements for continuous health related data like heart rate, respiration and galvanic skin responses in games. We have put a fourth a deep learning model which is trained on dynamic biometric signals to make the compressions to change according to the signals activity level and what is going on in the game. A simple neural encoder decoder architecture we present also meets the very low latency requirements of health related in game apps. We report that the results of our study support the that which we put forth as it reports real time health feedback parameters of the game environment. ession a machine learning based technique that adaptively learns how to compress and decompress data based on changing patterns provides an interesting alternative in contrast to static compression methods, neural models are capable of learning temporal relations-hips and preferencing essential features retaining higher resource efficincy without sacrificing clinical or gameplay meaningful this work investingates the use of adaptive neural networks for the compression of real time biometric information in the context of a health game we are concerned with the design and analysis of AI based compression techniques that adapt to signal properties and gameplay requirements Through testing with live streamed vital signs we seek to show the capability of neural compression to balance system performance against functional biometric input value our research provides the foundation for more inteligent, scalable health games that will operate well on various devices and network conditions.
Keywords:
